mobile.att.net MX meltdown
kai at pac-rim.net
Fri May 18 17:59:06 UTC 2001
In case you, like me, have a large number of email->paging forwards
in place for group-paging and other operational and commercial
purposes and were suddenly surprised today by chunks of 300/minute
warn/undeliverable msgs for @mobile.att.net addresses:
Apparently their MX (mta01.cdpd.airdata.com) had a little meltdown a little
while ago: from 12:18pm to 1:05p EDT, the MX refused connections entirely.
Whoops, and now that I write this, both NS's are off the air as
well. Should I hardcode my own MX entry into a local zone to
overcome this? I am getting tired of 300 mail warnings every 15 minutes.
Ouch. Ouch. Ouch.
[rant mode on]
Millions without working paging on their AT&T phones. For a little
Not that it ever works terribly predictable when it's actually up
either, but it's gotten better lately. Instead of paging delivery
times ranging from 10s to 6hrs, it's only been 10s to 30m for me
in recent months. With SMTP transport times over 1000's of pages
consistently under one minute. AT&T reps: "It's the Internet's fault"
A single-homed MX, even if there's a load-balancer there, and
2 nameservers, all in what appears to be the same announced
/24 from a legacy McCaw AS, makes you think how much "importance"
some incompetent beancounter at ATT WS has assigned to this
loss-leader of a profit-center within their wireless range of
Translate: because we give it to (nearly all?) our
customers for free, we don't have to care about how well
it works (the same could be said about their crappy, overloaded
CDPD network, which is seemingly back-hauling every last CDPD
frame in the country to washington state (via FR?) before
de-capsulating it and letting those IP frames lose. Ever
wondered about those 600ms ping times and 5 kbps effective
transfer rates and why "free unlimited Internet" wouldn't
entice you or me to ever pay for such a "performance" ?
(let alone them charging you at a rate that would make
every 900-number operator blush about the $/min flow.
if you hook up your PC via a 'tethered' connection as
described in the phone's manual, you'll discover after
the next billing cycle that they've raped you for $0.05/KB,
(e.g. up to $5.70/minute) conveniently hidden in the VERY
small fine print on their website, for making use of their
"free unlimited Internet". that is false and misleading
advertising at the least, and outright bait&switch fraud
at the worst - but hey, "very few customers are using
this functionality", "there is no support for this type of
operation", "I know nothing about this", "It's just in the
manual because we couldn't tell Mitsubishi to remove it
before printing them." You get the idea.
Add to that that I am on my 3rd Mitsubishi T250 phone now (
in 10 months, I have stopped requesting new phones a while
ago), the first one had ridiculously strong far-side echo for
all my callers, and the latter 2 hang almost everytime you
switch it into CDPD mode, except if you do it rather soon
after power-cycling it (read: taking the battery out, due to
the hang). And that was the early-adopter top-of-the-line
model the most money could buy with AT&T last summer, gee.
Why does the wireless data market not take off the way we
thought it would, uh oh? boy, do I have news for my bosses,
Omnisky, Aether, and Palm: lets get the hell out of this
[rant mode off]
"Just say No" to Spam Kai Schlichting
New York, Palo Alto, You name it Sophisticated Technical Peon
Kai's SpamShield <tm> is FREE! http://www.SpamShield.org
More information about the NANOG